It is well-known that work can impact health. In line with this, an individual’s state of health has been seen in occupational health psychology as multi-determined and as a function of various different factors such as stress at work or personality structure (Shirom 2003). Hereby, demands-resources-theory postulates that stress at work is contingent upon levels of working characteristics (Bakker and Demerouti 2007). On the one hand, job demands are supposed to result in negative health-related consequences. On the other hand, high levels of job resources are assumed to be health-protective (Bonde 2008; Luchman and González-Morales 2013; Stansfeld and Candy 2006).

However, the question of what makes working characteristics either stressful or helpful could vary between individuals. First, individuals differ in the perception of working characteristics and the appraisals they make (Lazarus and Folkman 1986; Semmer 2003). For example, several authors argue that individual-level factors such as locus of control, self-efficacy, self-esteem and neuroticism could explain additional variance above working conditions in predicting job strain (Semmer 2003). Moreover, they are supposed to moderate relationships between working characteristics and stress-related health consequences (Chang et al. 2012). Correspondingly, a growing body of literature focusing on associations between individual-level factors and health has to be recognized (for meta-analyses and reviews, see Alarcon 2011; Bagby et al. 2008; Kotov et al. 2010; Paterniti et al. 2002; Zellars et al. 2000). Results of these previous studies lead to the conclusion that individuals fundamentally differ in their response to stressful events at work. Thus, person-environment relationships are theoretically valuable in order to fully understand relationships between working characteristics and health.

Second, the presence of job demands and resources could also vary between the positions individuals have at work. For example, managers are supposed to face higher job demands but also higher job resources than employees (Skakon et al. 2011). Surprisingly, this target group has received only little attention in this specific field of research. Managers are not only important for the health of their assigned employees but also, from an economic point of view, with respect to the companies they are working for (Gregersen et al. 2011; Zimber et al. 2015). Moreover, a lack of person-related resources, appraisals and maladaptive behaviors could increase the vulnerability of individuals facing high levels of job demands such as managers do (Semmer 2003). Thus, it seems beneficial to examine relationships between job demands, resources, individual-level factors and health among managers.

Stress at work could lead to various negative health related drawbacks (Shirom 2003). Specifically depression may be influenced by psychosocial factors in the work place and is important not only because of its chronicity but also since people suffer from it frequently (Theorell et al. 2015). Depression, characterized by episodes of reduced mood and interest that persist for at least 14 days, is a prevailing cause of psychiatric morbidity and is therefore of great importance for society, research and practice. Moreover, increasing prevalence rates of depression have to be recognized (Andersen et al. 2011; Bonde 2008; Paterniti et al. 2002). Likewise, the etiology of depression has been viewed in clinical psychology as determined by multiple factors (e.g. working characteristics and personality) (Hammen 2005). Indeed, there is meta-analytic evidence that high levels of job demands could predict depression whereas resources could buffer against harmful working conditions (Bonde 2008; Nieuwenhuijsen et al. 2010; Theorell et al. 2015). Moreover, job resources such as feedback and decision latitude are supposed to improve work engagement, a state of vigor, absorption and dedication, which is linked to a variety of positive outcomes such as high job performance (Bakker 2011; Bakker et al. 2008).

As a logical consequence, it could be beneficial for managers to have a large reservoir of person-related resources to deal better with the highly demanding working conditions they are facing. On the other hand, individual aspects of a person also interplay with job resources and work engagement (Albrecht 2010; Bakker 2011). Work engagement can be seen as a combination of high work pleasure (dedication) with high activation (vigor and absorption) and is positively associated with job performance and several specific facets of it such as in-role performance, creativity and organizational citizenship behavior (Bakker 2011). Indeed, these are important outcomes especially for managers.

Particularly, the higher order construct core self-evaluations (CSE), which was originally introduced by Judge et al. (2003) refers to a basic, fundamental appraisal of one’s worthiness, effectiveness, and capability as a person, is supposed to function as both, a health preventive and a motivational personal resource (Albrecht 2010; Best et al. 2005; Doorn and Hülsheger 2015; Judge et al. 2004; Kammeyer-Mueller et al. 2009; Tsaousis et al. 2007). CSE is a broad higher order personality construct which is constituted by the shared variance of four well-known traits, namely (a) self-esteem, (b) generalized self-efficacy, (c) locus of control, and (d) neuroticism (Stumpp et al. 2010; Tsaousis et al. 2007). Judge et al. (2004) point out that individuals with high level of CSE “appraise themselves in a consistently positive manner across situations; such individuals see themselves as capable, worthy, and in control of their lives” (pp. 326–327). However, CSE has been examined mostly within the organizational context (Bono and Judge 2003; Judge et al. 2004; Tsaousis et al. 2007). Nevertheless, interest also increased in other fields of research and first attempts have been made to test whether CSE is associated with working characteristics and health (Chang et al. 2012).

Taking previous study results together, the evidence suggests that CSE is a construct, which appears to be related to job demands, resources, health and motivation. However, a broad empirical basis of studies examining associations between the full range of working characteristics, core self-evaluation and health is still lacking. Thus, it still remains unclear what role core self-evaluations play and whether the concept could explain additional variance. As a result, the present study adds to the contemporary literature and has two aims. First, the present study tests whether core self-evaluations may moderate the relationships between specific job demands, resources, depression and work engagement among a sample of managers. Second, the study examines whether core self-evaluation could explain additional variance in predicting depression and work engagement after controlling for job demands and resources, respectively.

In doing so, we contribute to the literature in several ways. In theoretical and empirical terms, we add the existing literature by examining the moderating effects and incremental validity of core self-evaluations. Although previous research examined relationships between working characteristics, the nature of the interplay between CSE and job characteristics remains unclear (Chang et al. 2012). Moreover, we test our hypotheses on a sample of managers with presumably relatively high levels of job demands and resources (Skakon et al. 2011; Zimber et al. 2015). Therefore, we go one step further since more homogeneous samples with rather low job demands have been used in most previous studies (Doorn and Hülsheger 2015). Although managers are an important target group for occupational health psychology, they have not been in the focus of previous studies (Mohr and Wolfram 2010). In practical terms, our results may be helpful in directing prevention efforts (Kotov et al. 2010). Furthermore, based on our results, better predictions could be made with regard to health related outcomes, which could be a valuable factor for personal assessment and development (Judge et al. 1998, 2002).

Theoretical Background

The Job Demands-Resources Model (JD-R)

The Job Demands-Resources (JD-R) model (Bakker and Demerouti 2007) discriminates between job demands and job resources. Job demands, comprising all physical, psychological, social or organizational aspects of the job, require prolonged effort and may therefore lead to physiological and psychological costs. Examples are high work intensity, work home conflicts, role conflicts and emotionally demanding actions with subordinates or clients (Bakker et al. 2010b). Job demands are not necessarily negative but they may turn into stress factors when personal resources are depleted and cannot be adequately recovered (Hockey 1997; Meijman and Mulder 1998; Safstrom and Hartig 2013; Siltaloppi et al. 2009). Job resources cover all physical, psychological, social or organizational aspects of the job that may contribute in achieving working goals, reducing job demands or stimulating personal development through a motivational process. Examples are feedback and social support from colleagues and supervisors, possibilities for development and decision latitude (Bakker and Demerouti 2007; Hackman and Oldham 1980; Xanthopoulou et al. 2007).

It is assumed that the combination of low job resources and high job demands could lead to health-related drawbacks due to a health impairment process. Indeed, several authors have shown that high job demands combined with low job resources lead to health related outcomes such as burnout (Alarcon 2011; Lee and Ashforth 1996; Schaufeli et al. 2008) and depression (Bonde 2008). On the contrary, job resources have been found to strengthen work engagement due to a motivational process (Bakker et al. 2008; Crawford et al. 2010; Hakanen et al. 2008).

Theoretical Incorporation of Personal Resources in the JD-R Model

It has been proposed in stress theory that personal resources such as self-esteem or locus of control can buffer the impact of stress on depression and therefore could contribute to better employee health (Lorant et al. 2003). Thus, attempts have been made to theoretically incorporate personal resources into the JD-R model (Hakanen and Schaufeli 2012; Hakanen et al. 2008; Xanthopoulou et al. 2007, 2009). In line with this, COR-theory is often used to explain the role of personal resources in the relationship between job demands, job resources and health. Accordingly, COR-theory (Hobfoll 2001) makes two core assumptions: First, people strive to obtain, retain, protect, and foster their resources. Secondly, resources are linked to other resources, which could lead to a gain spiral of resources. Conversely, facing a lack of resources is supposed to lead to a resource depletion process (loss cycle). Due to the investment of resources, people try to deal with threatening conditions and prevent themselves from negative outcomes (Xanthopoulou et al. 2007). In addition, COR theory suggests that resources become more important under demanding conditions (Hobfoll 2002). Having personal resources available in demanding environments may therefore be health-protective for managers. Thus, the combination of high job demands and individuals high in personal resources may result in higher levels of health than among those low in personal resources. Indeed, personal resources have been found to relate to job demands, job resources and health. Thus, personal resources are considered to be important for managers’ health (Alarcon et al. 2009; Christensen and Kessing 2006; Kim et al. 2009).

Job Demands as Drivers of Depression

In terms of job demands, perceived work intensity has been consistently found to be strongly associated with depression (Bonde 2008; Netterström et al. 2008; Theorell et al. 2015). In several studies, high workload has also been found to be associated with work-family-conflict. Work-family-conflict can be defined as an interrole conflict in which general demands of the job interfere with performing family-related responsibilities and activities (Demerouti et al. 2004; Ilies et al. 2007; Netemeyer et al. 1996). Meta-analytic results show that work-family-conflict is associated with depression as well (Amstad et al. 2011). Moreover, most jobs contain different task requirements and responsibilities, which may be especially valid for managerial jobs. As a result from multiple task requirements or responsibilities, role conflicts at work could occur when conflicting demands at the job arise (Alarcon 2011; Fried et al. 2008). However, recent meta-analytic results show only limited evidence for relationships between qualitative demands such as role conflicts and emotional demands and depression indicating that more research is needed in this regard (Theorell et al. 2015).

Based on the JDR-model and the outlined empirical evidence, we hypothesize:

  • H1a: Job demands (work intensity, emotional demands, work-home-conflicts, role conflicts) are positively associated with depression.

Relationships Between CSE and Depression

Prolonged exposure to job demands may also deplete one’s personal resources, which in turn may lead to higher levels of depression (Bakker et al. 2010a; Xanthopoulou et al. 2007, 2009). Previous research suggests that job demands may be differentially related to strain depending on whether a sufficient reservoir of personal resources such as positive core self-evaluations exist (Brunborg 2008; Doorn and Hülsheger 2015; Kammeyer-Mueller et al. 2009).

In a meta-analytic study, Kammeyer-Mueller et al. (2009) found that the four core traits of CSE were significantly associated with the use of coping strategies: CSE was positively associated with problem-solving coping and negatively with avoidance coping. In consequence, CSE may buffer the job demands-strain relationship due to a promotion of adaptive coping mechanisms. Thus, CSE may influence the way people deal with harmful working conditions and additionally affect employees’ reactivity to stressors. Due to a combination of a highly developed self-appraisal and confidence in their ability to cope with difficult situations, individuals with high levels of CSE may deal better with demanding work situations (e.g. high work intensity and emotional demands).

Moreover, studies revealed that those individuals with high levels of CSE strive for support, which enables them to deal effectively with multiple roles (Stumpp et al. 2009; Westring and Ryan 2010). Consequently, individuals high in CSE may perceive lower level of work-family-conflict and job-related role conflicts. According to COR-theory, high levels of CSE are supposed to reduce the depletion process of mental energy by providing a sense of self-worth and control while individuals facing high job demands (Chang et al. 2012; Doorn and Hülsheger 2015; Kammeyer-Mueller et al. 2009). Moreover, meta-analytic results of Chang et al. (2012) showed that CSE relates negatively to strain. As a result, the authors conclude that employees high in CSE appraise situations positively, which has an impact on both primary and secondary appraisals of the transactional stress model (Lazarus and Folkman 1986).

Based on the previous argumentation and the empirical evidence, we hypothesize:

  • H1b: CSE is negatively associated with depression.

Relationships Between Job Resources and Work Engagement

Based on job demands-resources theory, job resources drive work engagement due to their intrinsic and extrinsic motivational potential. In terms of the intrinsic potential, job resources satisfy basic human needs, which corresponds to the self-determination theory (Deci and Ryan 1985). For example, feedback is supposed to increase learning, which leads to more perceived job competence. In a similar vein, decision latitude satisfies the need for autonomy, whereas social support contributes to the need to belong. Furthermore, job resources may also evolve an extrinsic motivational potential, because they provide instrumental resources and enable one to dedicate and focus on the work task. Thus, in resourceful environments it is likely that a task can and will be completed successfully, which automatically leads to a higher goal attainment. Consequently, job resources such as social support from colleagues and feedback increase the likelihood of being successful in achieving one’s work goals. Whatever it may be, due to the satisfaction of basic needs or due to the achievement of work goals, it is likely that work engagement will be strengthened as a result (Bakker 2011). Empirical evidence showed that job resources such as social support from colleagues, feedback, decision latitude, and possibilities for development are positively associated with work engagement (Albrecht 2010; Bakker 2011).

Accordingly, we hypothesize

  • H2a: Job resources (decision latitude, possibilities for development, social support, feedback) are positively associated with work engagement.

Relationships Between CSE and Work Engagement

CSE is not only linked to depression, it is also supposed to be associated with motivational outcomes such as work engagement (Bakker 2011; Judge et al. 2004). From a theoretical point of view, individuals with high levels of CSE seem to be more committed to goal pursuit. As a result, those individuals are more likely to be intrinsically motivated due to higher levels of self-regard and goal self-concordance (Bakker 2011). As a consequence, individuals high in CSE focus more on the positive aspects of the task, which accordingly leads to higher motivation and work engagement (Chang et al. 2012). From an empirical point of view, meta-analytic results revealed, that positive relationships between CSE and motivation exist (Albrecht 2010; Bakker 2011; Chang et al. 2012).

Therefore, we hypothesize:

  • H2b: CSE is positively associated with work engagement.

CSE as a Buffer in Demand-Strain Relationships

High levels of CSE are supposed to enable individuals to cope more effectively with demanding situations by providing individuals with a sense of control. As a result, resources will be depleted to a lesser extent by demanding working conditions, which in agreement with the JDR-model leads to better health (Bakker and Demerouti 2007; Chang et al. 2012).

However, findings concerning moderating effects of CSE are mixed. For instance, whereas some studies demonstrated that CSE buffers the effects of social stressors and organizational constraints, other results showed effects in the opposite direction or failed to find support interactions between CSE and job demands (Chang et al. 2012). For example, Best et al. (2005) explored the role of CSE on burnout and found that core self-evaluations were negatively and moderately strong related to job burnout. Furthermore, the results of Tsaousis et al. (2007) revealed that CSE is a moderator in the relationship of two indicators of subjective well-being and physical functioning. However, contrary to their expectations, CSE did not moderate the relationship between subjective well-being and psychological health functioning. Doorn and Hülsheger (2015) expanded previous research and tested the idea that CSE acts as a buffer between job demands and negative strain reactions. Results showed that CSE moderated the relationship of job demands (emotional job demands, work load, and shift work) with psychological distress (irritation and depression).

According to this, we hypothesize:

  • H3a: CSE will negatively moderate the positive relationship between job demands (work intensity, emotional demands, work-home-conflicts, role conflicts) and depression, such as the relationship between demands and depression will be weaker among participants with high levels of CSE than among those with low levels of CSE.

CSE as an Amplifier in the Relationship Between Job Resources and Work Engagement

Individuals high in CSE believe that they are in control of their own environment. Consequently, it seems plausible that they will likely perceive higher decision latitude at work or try to proactively increase that resource (Bakker 2011; Judge and Kammeyer-Mueller 2011; Stumpp et al. 2009). Moreover, individuals high in CSE are more confident and therefore may feel more capable to professionally develop themselves than those low in CSE. Thus, individuals high in CSE proactively seek for task complexity. As a consequence, individuals high in CSE may perceive and seek more possibilities for development compared to those low in CSE (Judge et al. 2000; Srivastava et al. 2010). Furthermore, individuals high in CSE may seek more feedback and support than those low in CSE. As a consequence, individuals high in CSE will perceive higher levels of feedback and support at their job, which could also be beneficial in motivational-related terms (Chang et al. 2012; Song et al. 2013; Stumpp et al. 2009).

Additionally, CSE is supposed to bias how employees appraise and perceive their environment meaning that the relationships of job characteristics with outcomes are contingent upon CSE. Thus, positive aspects of the work should be more salient to individuals high in CSE. Empirically, these assumptions have received support, although some studies failed to find support for an interaction between job characteristics and CSE or even found support for the opposite prediction (Chang et al. 2012). However, CSE is supposed to be associated with an ability to take advantage of beneficial circumstances. More specifically, evidence suggest that individuals high in CSE are better at identifying and pursuing opportunities due to a specific motivational focused orientation. This is consistent with the assumption that individuals with higher levels of CSE view their environment and experiences more positively and are less sensitive to negative information compared with those individuals low in CSE (Chang et al. 2012).

Therefore, we hypothesize:

  • H3b: CSE will positively moderate the positive relationship between job resources (decision latitude, possibilities for development, social support, feedback) and work engagement, such as the relationship between job resources and work engagement will be stronger among participants with high levels of CSE than among those with low levels of CSE.

The Role of CSE in Explaining Depression

One could argue that individual-level factors do not equal the magnitude of work factors in explaining adverse health-related drawbacks such as exhaustion or depression (Knudsen et al. 2009; Melamed et al. 2006; Shirom 2005). However, work factors and their influence on health due to stress is closely related to the appraisal of stressors (Lazarus and Folkman 1986). Thus, working characteristics and their appraisals cannot hold for everyone in the same way. For example, when individuals are low in self-esteem they regard failure as “self-diagnostic” and appraise this as more stressful (Semmer 2003). In line with this, it should be noted that individuals high in CSE are supposed to appraise situations positively, which could impact the appraisals of stressors in general (Chang et al. 2012).

Moreover, individuals cope with stress in various ways. For example, an individual could try to cope with work overload by working overtime. As a result, this may in turn lead to fatigue or to problems with the family (Semmer 2003). As noted earlier, in a meta-analytic study Kammeyer-Mueller et al. (2009) found that CSE is positively associated with less avoidance coping, more problem-solving coping, and is not strongly interrelated with emotion-focused coping. In agreement with this, adaptive coping mechanisms among individuals high in CSE may influence the way in which they deal with the exposure to job demands. Thus, it seems plausible that CSE has the ability to contribute to explaining depression due to differing appraisals and behavior contingent upon the levels of CSE.

Therefore, we hypothesize:

  • H4a: Core self-evaluation will show incremental validity above job demands (work intensity, emotional demands, work-home-conflicts, role conflicts) in explaining depression among managers.

The Role of CSE in Explaining Work Engagement

Individuals high in CSE are supposed to be sensitive for positive information and to approach positive stimuli. In line with this, research has shown that high levels of CSE are correlated with a strong approach temperament and weak avoidance temperament (Chang et al. 2012). Thus, differences in approach/avoidance temperaments among individuals could influence the degree to which people focus on positive or negative information when evaluating situations (Chang et al. 2012). Therefore, individuals high in CSE may focus on the beneficial aspects of their job resources, which could drive their work engagement positively (Bakker 2011). As a result, CSE may also contribute in explaining work engagement.

Accordingly, we hypothesize:

  • H4b: Core self-evaluation will show incremental validity above job resources (decision latitude, possibilities for development, social support, feedback) in explaining work engagement among managers.

Methods

Participants and Procedure

Participants were recruited from different branches of industry (e.g. personal services sector, IT and Sciences services sector) in order to maximize variability in job demands and job resources. Study participants were recruited from training facilities of the social and health services (sample A), a leaders association (sample B) and a pharmaceutical and law firm (sample C). In all samples, questionnaires were distributed with a declaration of consent assuring anonymity. The participants in sample A sent their completed questionnaires directly to the research group in a prestamped envelope. In return for the participation in the study, we offered a summary report for each participant involved. 408 questionnaires were distributed and 161 were returned, yielding a response rate of 39.5 %. In sample B, 70 managers, who had subscribed an internal newsletter of a leaders association representing managers on the federal political scene in Germany and Europe, participated. Managers received an invitation via e-mail to fill in an online questionnaire. However, the calculation of a meaningful response rate was not possible in this case since no information was available regarding the number of active participants in the of survey panel and rather inactive participants who only read the results but do not actively participate by answering surveys. We compared the response rate with the response rates of other studies using this survey panel and found that the degree of participation was on average. Participants in sample C sent their completed questionnaires directly to the research group. Fifty-one managers of a pharmaceutical and a law firm participated, yielding a response rate of 65 %.

The final study sample consisted of N = 282 managers. Half of the participants were female (54.4 %). Managers were between 25 and 70 years old (M = 47.2). The sample was fairly well-educated and 51 % held a university degree. 95 % had managerial responsibility whereas 5 % had project, process or product responsibility. 15.79 % of the participants worked in a top management position (e.g. managing director or CEO). The majority of the sample held a mid-level management position as a divisional manager (32.39 %) or as a department manager (31.97 %). 19.83 % of the participants were group managers.

Forty-three percent were responsible for 1 to 19 employees, 21.4 % for 20 to 30 employees, 8.2 % for 31 to 40 employees and 20.6 % for more than 40 employees. Sixty-one percent of the participants worked full-time, and 39 % part-time. The total sample had fairly long weekly working hours (M = 47.07; SD = 8.0). The majority of the sample (60.1 %) worked in the personal services sector. The remaining managers worked in the commercial sector (15.7 %), IT- and Sciences services sector and other services (5.2 %). Sixty-two percent worked for private companies, 38 % in the public sector.

Measures

Depression was measured by the nine-item (e.g. “Little interest of pleasure in doing things“) German version of Patient Health Questionnaire (PHQ-9) (Löwe et al. 2002). Responses were rated on a 4-point frequency scale ranging from 0 (not at all) to 3 (every day). Cronbach’s α was .84. In terms of validity, confirmatory factor analysis using the R programme (R-Core-Team 2015) revealed acceptable fit to the data for the measure (χ 2 = 37.923, df = 24, p = .035; comparative fit index [CFI] = .981, Tucker-Lewis index [TLI] = .971, root mean square error of approximation [RMSEA] = .046, standardized root mean square residual [SRMR] = .031)

Work Engagement was assessed with the nine-item German version of the Utrecht Work Engagement Scale (UWES (Schaufeli et al. 2002)). The UWES scale reflects three underlying dimensions, which are measured with three items each: vigor (e.g., “At my work, I feel bursting with energy”), dedication (e.g., “My job inspires me”), and absorption (e.g., “I get carried away when I am working”). Answer categories ranged from 0 (not at all) to 6 (always / every day). Cronbach’s α was .94. In terms of validity, confirmatory factor analysis indicated satisfactory fit to the data (χ 2 = 78.251, df = 24, p = .00, CFI = .972, TLI = .959; RMSEA = .090, SRMR = .026).

Core Self Evaluation (CSE) was measured with the 12 item German version (Stumpp et al. 2010) (e.g. “I am confident to get the success I deserve in life”) of the CSE-scale (Judge et al. 2003). Items were scored on a five-point likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). Some items were recoded so that high scores indicate high level of CSE. Cronbach’s α was .85. In terms of validity, confirmatory factor analysis revealed acceptable fit to the data (χ 2 = 79.632, df = 45, p = .00, CFI = .966, TLI = .951, RMSEA: .052, SRMR: .038).

Job Demands

We used different types of job demands: quantitative (e.g., work intensity) and qualitative (e.g., emotional demands) demands. The German Fragebogen zum Erleben von Intensität und Tätigkeitsspielraum in der Arbeit (FIT) questionnaire (Richter et al. 2000) was used to assess work intensity (6 Items, e.g., “At this work, I have to do many things at the same time.”). Answer categories ranged from 1 (no) to 4 (yes). Cronbach’s α was .77. In terms of validity, confirmatory factor analysis for measure indicated acceptable fit to the data (χ 2 = 13.564, df = 9, p = .139, CFI = .989, TLI = .982, RMSEA = .042, SRMR = .025).

The German version (Nübling et al. 2005) of the Copenhagen Psychosocial Questionnaire (Kristensen and Borg 2000) was used to assess job demands such as emotional job demands (3 Items, e.g., “Is your work emotionally demanding?”), role conflicts (4 Items, e.g., “Are contradictory demands placed on you at work?”) and work-home-family conflicts (5 Items, e.g., “The demands of my work interfere with my home and family life.”). Answer categories ranged from 0 to 100, where 0 represents the minimum and 100 the maximum. Cronbach’s α were .78 for emotional demands, .92 for work-home-family-conflicts and .70 for role conflicts. In terms of validity, confirmatory factor analysis revealed acceptable fit to the data for the measures (χ 2 = 158.586, df = 50, p = .00, CFI = .933, TLI = .911, RMSEA = .088, SRMR = .077).

Job Resources

We used job resources from different levels: organizational level (possibilities for development), the task (decision latitude) and interpersonal relations (social support and feedback). The German “Fragebogen zum Erleben von Intensität und Tätigkeitsspielraum in der Arbeit (FIT)” questionnaire (Richter et al. 2000) was used to assess decision latitude (7 Items, e.g., “I can plan my work on my own”). Answer categories ranged from 1 (no) to 4 (yes). Cronbach’s α was .62. In terms of validity, confirmatory factor analysis revealed acceptable fit to the data (χ 2 = 13.930, df = 11, p = .237, CFI = .987, TLI = .976, RMSEA = .031, SRMR = .028).

The German version (Nübling et al. 2005) of the Copenhagen Psychosocial Questionnaire (Kristensen and Borg 2000) was used to assess job resources such as possibilities for development (4 Items, e.g., “Do you have the possibility of learning new things through your work?”), feedback (2 Items, e.g., “How often do you talk with your superior about how well you carry out your work?”) and social support from colleagues and supervisors (4 Items, e.g., “How often do you get help and support from your colleagues?”). Answer categories ranged from 0 to 100, where 0 represents the minimum and 100 the maximum. Cronbach’s α were .91 for possibilities for development, .83 for social support and .41 for feedback. Managers appear to be a group, which is not easy to come by. Therefore, we had to make sure that we measure all constructs economically and decided to incorporate a brief measure of feedback. In terms of validity, confirmatory factor analysis indicated satisfactory fit to the data for the measures (χ 2 = 106.919, df = 27, p = .00, CFI = .949, TLI = .915, RMSEA = .102, SRMR = .045).

Statistical Analysis

T-tests were used to compare depression scores of our managerial sample with a representative sample of the general population in Germany (Glaesmer et al. 2009; Lakens 2013).

To test the hypotheses that CSE moderates the positive relationships between job demands and depression as well as job resources and work engagement, we used hierarchical regression analysis. First, we centered all independent variables (Aiken and West 1991). The predicted two-way interactions were then tested in a series of eight separate hierarchical regression analyses for both, depression and work engagement. In each hierarchical regression, a specific job demand, or a specific job resource together with CSE were included in the first step of the regression equation. The interaction term between CSE and the specific job demand or job resource was included in the second step of the regression. Through this approach, we examined the extent to which the interaction term between a job demand and a job resource, together with CSE explained significantly more of the variance in depression and work engagement, after controlling for the main effects. Moreover, we tested for the incremental validity of CSE in explaining depression and work engagement. Therefore, we controlled in the first step of the regression for all job demands and resources. CSE was included in the second step of the regression. Thus, we examined the extent to which CSE explained significantly more of the variance in depression and work engagement, after controlling for the main effects of job demands and resources. Moreover, we adjusted p values for multiple comparisons with a Bonferroni correction. To determine the strength of the effects, we used f2 as recommended by Cohen (1988) who suggests f2 values of 0.02, 0.15, and 0.35 represent small, medium, and large effect sizes. We used R for statistical analysis (R-Core-Team 2015).

Control Variables

Due to gender differences associated with depressive symptoms, we decided to control for the sex of participants (Herrero et al. 2012). Moreover, we tested whether age was significantly associated with the dependent variables. However, age was not significantly associated with one of the dependent variables.

Results

Gender was significantly (t(272) = 3.41, p < .001) associated with one of the dependent variables (depression). Females suffered more frequently from depressive symptoms than males. However, age was not associated with the dependent variables. Therefore, gender functioned as a covariate in all regression analyses on depressive symptoms.

Descriptive statistics, zero-order correlations and Cronbach’s α are provided in Table 1.

Table 1 Descriptive statistics, zero-order correlations and Cronbachs α
Full size table

All constructs had satisfactory reliability (Cronbachs α), with only two exceptions for decision latitude (.62) and feedback (.41). In the latter case, however, it should be noticed that only two items constitute the scale. Correlations among the study variables were as expected and low to high in size. Depression and work engagement were negatively associated. Core self evaluations were positively associated with work engagement and negatively with depression. On the one hand, job demands related positively and strongly to depression and weakly to work engagement. On the other hand, job resources related positively and strongly to work engagement and were weakly associated with depression. Thus, zero-order correlations provided further evidence for the validity of the measures. The mean score for work engagement was on average, whereas the depression score (M = 5.81, SD = 4.05) in the present sample was significantly (t(2712) = 14.25, p < .001) higher compared with a representative sample of the German general population.

To test the hypotheses (H1a, H1b, H3a, H4a) that depression is a function of multiple job demands, and more specifically, whether CSE moderates the relationship between work intensity, emotional demands, work-home-family conflicts, role conflicts and depression, a hierarchical multiple regression analysis was conducted. Results are displayed in Table 2. The overall model was significant, R 2 = .390, F(6, 266) = 28.44, p < .001. Thus, H1a and H1b received support. Moreover, the incorporation of CSE, after controlling for all job demands, showed a significant increase in R 2 which represents a medium effect (f2 = 0.31).

Table 2 Depression regressed on job demands and core self-evaluations controlled for gender
Full size table

Therefore, H4a received support.

Subsequent simple slope analyses revealed significant slopes for the regression of depression on low CSE for emotional demands (β = .06, t(271) = 5.45, p < .0001), and on high CSE for emotional job demands (β = .03, t(271) = 1.67, p < .0001), which represents a large effect (f2 = 0.60). The supported simple slopes for depression were also significant at the p < .001 level, after Bonferroni correction for multiple comparisons. No significant interaction effect between CSE and other job demands was found. Thus, H3a is partially supported (see Fig. 1).

Fig. 1
figure 1

CSE moderates the relationship between emotional demands and depression

Full size image

To test the hypotheses (H2a, H2b, H3b, H4b) that work engagement is a function of multiple job resources, and more specifically whether CSE positively moderates the relationship between decision latitude, possibilities for development, social support, feedback and work engagement, a hierarchical multiple regression analysis was conducted. Results are displayed in Table 3. The overall model was significant, R 2 = .153, F(5, 267) = 9.363, p < .001. Thus, H2a and H2b received support. Moreover, the incorporation of CSE, after controlling for all job resources, showed a significant increase in R 2 which represents a small effect (f2 = 0.04). Thus, H4b received support.

Table 3 Work engagement regressed on job resources and core self-evaluations
Full size table

Simple slope analysis revealed that the slope for the regression of work engagement on possibilities for development was significant (p < .0001) for low CSE (β = .01, t(276) = 3.74). The supported simple slope for work engagement was also significant at the p < .001 level, after Bonferroni correction. For possibilities for development, the slope of high CSE was not significant (p > .05) meaning that there is no positive moderation of CSE. Therefore, the effect size represents a small effect (f2 = 0.14). No significant interaction effect between CSE and other job resources was found. Thus, H3b is rejected (see Fig. 2).

Fig. 2
figure 2

CSE moderates the relationship between possibilities for development and work engagement

Full size image

Discussion

The purpose of this study was to examine the role of CSE in the relationship between job demands and depression and in the relationship between job resources and work engagement, respectively. Therefore, a sample of managers presumably facing high job demands and resources was recruited to examine the relationships and test for interaction effects between demands, resources, CSE, depression and work engagement.

Based on the JDR-model, hierarchical regression analyses were performed. Results revealed that besides job demands and resources, CSE is able to explain additional variance of depression and work engagement. Thus, hypotheses H1a, H1b, H2a and H2b received support. Moreover, we focused on the moderating role of CSE in explaining depression and work engagement. In doing so, we are able to gain a better understanding regarding the influence of CSE on depression and work engagement. We found two significant interaction effects: Regression analysis revealed that CSE is able to buffer the positive relationship between emotional job demands and depression, such as managers high in CSE had lower depression scores compared with those low in CSE. Therefore, H3a received partial support. In addition, results suggest that in terms of work engagement, only managers low in CSE benefit from more possibilities for development. Thus, our hypothesis that CSE is positively moderating the relationship between job resources and work engagement (H3b) has to be rejected. Furthermore, we hypothesized that CSE will have incremental validity in explaining depression and work engagement. Indeed, CSE significantly increased the explanation rate in predicting depression as well as work engagement and showed incremental validity above the working conditions. Thus, hypotheses H4a and H4b received support.

The present study adds to existing literature since researchers have previously failed to find empirical support for a buffering effect of CSE (Kammeyer-Mueller et al. 2009). Since we investigated more job demands and additionally tested for the effects of CSE in the relationship between work engagement and job resources, we replicate and expand previous research (Doorn and Hülsheger 2015) which was able to find an interaction effect of CSE in the relationship between job demands and health. Whereas we could not find a moderating effect for work intensity, we found a buffering effect for emotional job demands which is in line with the findings of Doorn and Hülsheger (2015) and provides further insight for an existing relationship between qualitative demands such as emotional demands and depression (Theorell et al. 2015). Moreover, we expand previous research since our managerial sample had relatively high job demands whereas previous research examined relationships on samples facing rather low job demands (Doorn and Hülsheger 2015).

Furthermore, we found that CSE moderates the relationship between possibilities for development and work engagement. Individuals low in CSE perceived low work engagement when just few possibilities for development were offered. However, more possibilities for development offered could boost work engagement in individuals low in CSE. Therefore, our finding contradicts the suggestions that CSE biases how employees appraise the environment, such that positive aspects of work are thought to be more salient to individuals high in CSE (Chang et al. 2012; Judge et al. 1997). Thus, CSE may either have a direct effect on work engagement such that positive self-views spill over or an indirect effect by influencing the actions in which the individuals engage (e.g., persisting on tasks) (Chang et al. 2012). This finding is in agreement with results of Ng and Feldman (2010) who also found support for the prediction that individuals low in CSE benefitted more from favorable working conditions than those high in CSE. Accordingly, the interplay and working mechanisms between CSE and job resources in explaining work engagement still remains unclear.

However, this result may shed more light on the question whether CSE is a moderator due to mobilizing new resources or due to minimizing resource loss (Doorn and Hülsheger 2015). The potential benefit of CSE to buffer negative-effects of job demands or to act as an amplifier of job resources on work engagement has been based on the COR-theory. Theoretically speaking, individuals may differ in how they actively seek to increase their resources, but may also vary in attempting to minimize resource loss. It could be that individuals high in CSE are more efficient in activating and recruiting job resources, or in managing existing resources (Doorn and Hülsheger 2015; Judge and Hurst 2007). Our results in terms of work engagement provide evidence for the latter case.

The question why we could not find an interaction effect between CSE and the other job demands and resources is hard to answer. This may be due to methodological issues since regression analysis is a rather conservative approach in detecting moderating effects (Cohen and Edwards 1989; Semmer 2003).

Limitations and Suggestions for Future Research

Our results clearly add to existing literature since they show that CSE relates to both, health and motivation (Chang et al. 2012; Judge et al. 2004). Although our results broaden the existing knowledge about the role of CSE and its potential motivational and health-related effects, the study has several shortcomings that should be acknowledged. Obviously, our research relies on self-reported data from managers. Therefore, our results do not allow a clear cause-effect relationship and could indeed be biased due to common method bias. However, we hypothesized and analyzed quite specific interactions with a cross-sectional design, which is already a conservative approach in moderated regression analysis (Podsakoff et al. 2012; Siemsen et al. 2010). Therefore, we conclude that our results are not easily attributable to third factors, reverse causation or common method variance. Furthermore, we adjusted p values with a Bonferroni correction. Although this is an extremely conservative approach and could lead to a high rate of false negatives, the results revealed that supported simple slopes were also significant after Bonferroni correction. This provides further evidence for the interaction since potential effects of chance can be ruled out.

Although managerial samples are generally quite hard to recruit, more studies should use longitudinal designs in order to address issues concerning causal inferences. Moreover, the theoretical framework that explains how CSE influences hypothesized outcomes has to be further evaluated (Chang et al. 2012). For example, based on the results, one could conclude that the motivational effect of CSE on work engagement is better explained by a direct effect of CSE than due to influencing cognitions or appraisals. This direct effect could be working either through a process of emotional generalization or by influencing actions. Therefore, more research on the working mechanisms of CSE is needed (Chang et al. 2012).

Moreover, it should be discussed whether the low reliability of decision latitude and feedback has affected the results. Estimates of the regression coefficients in multiple regression analysis with unreliable variables are biased and lead also to a low reliability of the product term. As a result, the low reliability of the product term in moderated regression analysis even amplifies this bias. Consequently, this leads to a deflation of the interaction effect. Moreover, a loss of the statistical power for testing the interaction hypothesis is expectable.

Our study clearly revealed that managers had significantly higher depression scores compared to the general population in Germany. Although managers play a key role in the organizational context, their health has received less attention so far. Previous research focused more extensively on health among subordinates (Zimber et al. 2015). Representative samples with managers from different branches are needed to evaluate the demand for health promotion more accurately for this important target group.

Implications for Theory and Practice

Our results are interesting in theoretical and practical terms. In theoretical terms, our results show that CSE is related to both, health and motivation. Thus, CSE plays a dual role. On the one hand, CSE could buffer negative health-related consequences. On the other hand, CSE fosters motivation. In line with this, more possibilities for development help to compensate for a lack of motivation among individuals low in CSE. Therefore, personnel development measures may be particularly important for those people who do not see themselves as capable, worthy, and in control of their lives (Judge et al. 2004). On the contrary, those who appraise themselves positively across situations may not benefit from more possibilities for development in terms of enhancing their work engagement since they may feel engaged per se. In practical terms, it could therefore be beneficial to assess CSE in personnel selection for highly emotional demanding jobs since CSE could buffer negative strain reactions. Individual level factors such as CSE are relatively stable constructs and only slightly malleable due to trainings. Thus, employees and managers facing highly emotional demands should be stocked with a strong reactivity to stressors.